72 



of the Royal Artillery stationed at the Observatory (7 '"9 3) differs so 

 slightly from that derived from the observations made at the anchor- 

 age from the earliest period at which observations are recorded 

 (i. e. 8' "05), that we may practically regard them as the same. To 

 examine whether this has been a uniform rate throughout the 236 

 years, or otherwise, the same calculation which gives 8''05 as the 

 most probable average rate of change between 1610 and 1846, will 

 give also for each of the years in which the Declination was observed 

 the most probable values of the Declination corresponding to the 

 same rate of change supposed uniform. These calculated values are 

 placed in the Table opposite to the years to which each belongs, and 

 adjoining the observed values. The differences are shown in the 

 next column. On inspecting these, we perceive that not one of the 

 differences exceeds the limits, which, with a due consideration of the 

 irregularities to which magnetic observations made on board ship are 

 liable, may be ascribed to accidents of observation ; and, what is still 

 more important, that they fall indiscriminately to the east and to the 

 west of the values calculated on the supposition of a uniform rate, 

 and without the slightest appearance of any systematic character 

 which might indicate that the rate had been otherwise than regular. 

 We have reason to conclude, therefore, that, from the earliest date to 

 which we can refer, the progression of secular change at St. Helena 

 has gone on from year to year, as nearly as may be, in one uniform 

 annual rate. 



" The instruction to be derived from the St. Helena observations 

 does not however stop here. By a suitable arrangement of the 

 observations of the eight years, they may be made to show that, 

 when allowance is made for comparatively very small irregularities 

 superimposed upon the regular march of the phenomenon by dis- 

 turbing causes which will be treated of in the sequel, the average 

 annual change takes place by equal aliquot portions in each month 

 of the year. The eight years of observation commenced with June 

 1841 : if we take a mean of the eight monthly means in the eight 

 Junes from ] 841 to 1848, we shall have a better assured mean value 

 of the Declination corresponding to the month of June, than if we 

 had confined ourselves to a single year. If we then do the same 

 with the eight Julys, and with each of the other months in succes- 

 sion, we shall have twelve monthly values for a year commencing 



